That Real Estate Tech Guy

Host argues companies must redesign work architecture before calculating AI economics to avoid operational failure.

22 min Listen to the original
AI EconomicsBusiness Process RedesignOperational ArchitectureHeadcount StrategyWorkflow Design

Summary

The host argues that starting with AI economics is a costly mistake because it assumes the current business structure is static. He illustrates that cutting headcount without redesigning workflows leads to context loss and employee burnout. The speaker warns that AI can amplify constraints if downstream systems cannot process increased volume. He proposes a framework where work definition and responsibility assignment precede economic calculations. This approach aims to decouple growth from headcount increases by redesigning infrastructure first.

Episode sections

  1. 0:01 – 0:40

    Introduction and Core Thesis

    Host states that starting with AI economics is a major mistake and companies should not begin by calculating cost savings based on current AI capabilities.

  2. 0:40 – 2:07

    The Flaw of the Economics-First Approach

    Speaker explains that looking at current payroll ignores 'invisible work' and infrastructure responsibilities, leading to broken processes when headcount is cut.

  3. 2:07 – 7:30

    Case Study: Customer Success Team Reduction

    Illustrates that reducing a 10-person team to three without redesign causes context loss, exception pile-ups, and burnout.

  4. 7:30 – 8:36

    AI as a Constraint Amplifier

    Warns that focusing on cost-cutting ignores capacity; AI generating more leads creates pressure on downstream constraints if the organization cannot process them.

  5. 8:36 – 11:36

    The AI Labor Architecture Framework

    Outlines a framework where economic architecture is an outcome, advising to start with work definition, responsibility assignment, and infrastructure design.

  6. 11:36 – 14:53

    Redesigning Work Before Calculating Savings

    Argues that understanding actual work and removing inefficiencies first allows for 5x capacity growth rather than just headcount reduction.

  7. 14:53 – 19:46

    Changing the Growth-Headcount Relationship

    States that AI can break the historical link between growth and linear headcount increases if context and handoffs are managed by systems.

  8. 19:46 – 22:15

    Conclusion and Actionable Advice

    Advises listeners to set aside initial ROI calculations, redesign business operations first, and then measure the new economics.

Claims worth verifying

Jordan Samuel Fleming is the CEO of Smartphones.

Unidentified speaker ·0:32

How to verify: Check corporate filings or the official website for Smartphones to identify the CEO.

Smartphones is a phone system built for real estate investors that connects to real estate investing CRMs.

Unidentified speaker ·10:07

How to verify: Visit the Smartphones website or review marketing materials for the product's features and target audience.

AI agents can communicate, answer questions, monitor behavior, follow up, and maintain context across multiple relationships 24 hours a day.

Unidentified speaker ·2:16

How to verify: Review technical documentation or capability lists from major AI agent providers (e.g., OpenAI, Anthropic, or specialized AI agent platforms).

AI labor has the potential to fundamentally change the economics of a business.

Unidentified speaker ·1:32

How to verify: Review industry reports on AI adoption in enterprise settings, such as those from McKinsey, Gartner, or Deloitte, regarding productivity and cost structures.

If an AI sales agent produces 10x more qualified opportunities, it may create 10x the pressure on the next constraint if the organization cannot process them.

Unidentified speaker ·8:12

How to verify: Apply Theory of Constraints principles to business case studies or operational management literature to verify the impact of bottleneck amplification.

Technology has improved productivity enormously, but in many parts of a business, growth still creates a labor demand.

Unidentified speaker ·16:24

How to verify: Review economic data on labor productivity vs. employment growth in service sectors, such as data from the Bureau of Labor Statistics.

Do your own research

  • Verify the speaker's credentials and any conflicts of interest regarding AI consulting or software sales.
  • Review case studies cited for specific metrics on context loss and burnout rates in reduced teams.
  • Check industry reports on the correlation between AI lead generation and downstream conversion bottlenecks.
  • Analyze competitor implementations of AI labor architecture to compare outcomes with the speaker's framework.
  • Research academic or industry literature on 'invisible work' and its quantifiable impact on operational efficiency.
  • Look for counter-arguments from traditional ROI-focused AI implementation guides.
  • Examine the speaker's proposed framework for measurable KPIs beyond headcount reduction.
  • Verify the historical data cited on the linear relationship between growth and headcount in the relevant industry.
Full transcriptexpand
0:01
Unidentified speaker: I can tell you one thing.
0:03
Unidentified speaker: Technology is the single most important aspect of every business that is successfully scaled.
0:10
Unidentified speaker: It's time for that real estate tech guide.
0:12
Unidentified speaker: It's your weekly test to explore how technology can help your real estate business explode.
0:18
Unidentified speaker: Each week you'll hear from real estate investors who have been there and done that and find out their favorite technology test.
0:25
Unidentified speaker: Listen is Jordan Street's With Tech Companies to learn about new technologies and new ideas that will help you scale your business.
0:32
Unidentified speaker: And now, join your host, Jordan Samuel Fleming, CEO of Smartphones, for this week's episode.
0:40
Unidentified speaker: I think one of the biggest mistakes companies are gonna make with AI is also one of the easiest mistakes to understand.
0:48
Unidentified speaker: They're gonna start with the economics.
0:52
Unidentified speaker: They're gonna look at what AI agents can do, look at what they currently spend on people, and immediately start doing the math.
1:01
Unidentified speaker: We have 20 people doing this today, could we do it with 10?
1:04
Unidentified speaker: Could we do it with five?
1:07
Unidentified speaker: Could AI agents handle three times the volume without us hiring anybody else?
1:12
Unidentified speaker: And I completely understand why.
1:15
Unidentified speaker: If you're running a business, of course you care about the economics.
1:19
Unidentified speaker: I do.
1:20
Unidentified speaker: Payroll matters, margins matter, capacity matters, revenue matters.
1:27
Unidentified speaker: The amount of work you can produce for every dollar you spend matters.
1:32
Unidentified speaker: And there is absolutely no question in my mind that AI labor has the potential to fundamentally change the economics of a business.
1:43
Unidentified speaker: But I think there is a problem if you start the conversation with economics.
1:49
Unidentified speaker: See, if you start there, there's a really good chance you're starting in the wrong place.
1:56
Unidentified speaker: Because if you do this, you're trying to calculate the economic impact of AI before you've redesigned the business to actually take advantage of it.
2:07
Unidentified speaker: Now, we've talked about customer success.
2:09
Unidentified speaker: Imagine you have a customer success team of 10 people, and you look at what AI agents can do today.
2:16
Unidentified speaker: I mean, they can communicate, they can answer questions, they can monitor, they can recognize changes in behavior and follow up, and they can maintain context across multiple relationships and exercise judgment within their defined responsibility 24 hours a day.
2:35
Unidentified speaker: They can potentially manage volumes of customer interactions that would be impossible for one person or one team to manage.
2:44
Unidentified speaker: So it makes sense.
2:46
Unidentified speaker: Somebody looks at the customer success team and says, Well, we got 10 people doing this.
2:52
Unidentified speaker: With AI, we could probably do it with three.
2:56
Unidentified speaker: Maybe.
2:57
Unidentified speaker: Maybe you can.
2:59
Unidentified speaker: But how did we get to three?
3:00
Unidentified speaker: What did we actually analyze?
3:03
Unidentified speaker: Did we understand the work?
3:06
Unidentified speaker: Did we understand the responsibilities those 10 people actually carry?
3:11
Unidentified speaker: Did we understand which parts of their work are visible and which parts aren't?
3:16
Unidentified speaker: Did we understand the judgment they're actually exercising and the exceptions they're handling, the information they're carrying, the relationships they're managing?
3:27
Unidentified speaker: Did we understand how work arrives to them and how it moves between them?
3:33
Unidentified speaker: And what happens when another department needs to become involved?
3:38
Unidentified speaker: Did we understand which of those people are actually functioning as part of the infrastructure of the business?
3:46
Unidentified speaker: Or did we just look at 10 salaries and start subtracting?
3:52
Unidentified speaker: Because those are very different exercises.
3:55
Unidentified speaker: And this is where I think the economics first approach can get you into trouble.
4:00
Unidentified speaker: So let's say you go ahead and you introduce AI agents, and you reduce the customer success team from 10 people to three.
4:09
Unidentified speaker: Initially, well, the economics will look and feel fantastic.
4:13
Unidentified speaker: Payroll comes down.
4:15
Unidentified speaker: Cost per customer comes down.
4:17
Unidentified speaker: Maybe response times improve.
4:20
Unidentified speaker: Your agents can handle enormous volumes of communication, so your spreadsheets probably look great.
4:27
Unidentified speaker: But then things could start happening.
4:30
Unidentified speaker: An agent recognized that a customer needs something from finance, but that work doesn't move properly.
4:37
Unidentified speaker: Or one of the remaining people then has to step in.
4:41
Unidentified speaker: Or a customer has an unusual commercial arrangement.
4:45
Unidentified speaker: But the context surrounding that arrangement was historically carried by someone who's no longer there.
4:52
Unidentified speaker: Now somebody has to reconstruct it.
4:55
Unidentified speaker: The agents are generating and responding to far more activity than the old team could, but your existing systems and processes can't absorb that volume, and exceptions start piling up.
5:07
Unidentified speaker: One of the three remaining people, they probably become the person who knows how everything works.
5:15
Unidentified speaker: She starts checking, chasing, rooting, explaining, recovering context, making sure the work keeps moving, and suddenly she's overwhelmed.
5:24
Unidentified speaker: Now, now you're looking at the business thinking, wait, I thought AI was supposed to give us leverage.
5:32
Unidentified speaker: It did.
5:34
Unidentified speaker: You just didn't redesign the architecture to actually take advantage of that leverage.
5:41
Unidentified speaker: You started with the economic outcome you wanted, and you tried to force the existing businesses and the business structure to produce it.
5:52
Unidentified speaker: Now, there is a lot of that going on right now because AI makes the economic argument really, really tempting.
6:00
Unidentified speaker: When you see an AI agent capable of executing work at a scale that previously required a team of people, it's almost impossible not to immediately calculate the saving.
6:13
Unidentified speaker: But there's a big assumption that goes inside the heart of that calculation.
6:18
Unidentified speaker: You're assuming that the existing organization is simply a collection of work being performed by people.
6:27
Unidentified speaker: So if AI agents can perform that work more economically, you can substitute one form of labor for another and collect the difference.
6:36
Unidentified speaker: But if you've followed this series, you probably know by now, I don't think businesses are that simple.
6:42
Unidentified speaker: Those people aren't simply executing the activities in their job descriptions.
6:48
Unidentified speaker: They're carrying responsibilities, they're making decisions, they're maintaining relationships, they're holding context.
6:58
Unidentified speaker: No, with all of that work, moving work between systems and noticing when something happens, resolving exceptions, they're doing that invisible work.
7:08
Unidentified speaker: And it means they're part of the infrastructure that lets the organization function.
7:14
Unidentified speaker: So if your AI strategy begins with where can we reduce the number of people, you may be making an economic decision before you understand what those people are actually doing.
7:27
Unidentified speaker: And that's backwards.
7:30
Unidentified speaker: And the same thing applies at the other end of the equation.
7:34
Unidentified speaker: Because economics first doesn't only mean cost cunning.
7:38
Unidentified speaker: Maybe you're not trying to reduce headcount.
7:40
Unidentified speaker: Maybe you're looking at AI and thinking, man, this is amazing.
7:44
Unidentified speaker: We can do 10 times more.
7:47
Unidentified speaker: Okay.
7:48
Unidentified speaker: Can the business actually absorb 10 times more?
7:53
Unidentified speaker: We talked about this in the very first video in this series.
7:56
Unidentified speaker: If your AI sales agent can produce 10 times more qualified opportunities, but the rest of the organization can't process those opportunities in time, you haven't necessarily created 10 times the economic value.
8:12
Unidentified speaker: You may have created 10 times the pressure on the next constraint.
8:17
Unidentified speaker: And again, the capability is real.
8:20
Unidentified speaker: I don't doubt that.
8:22
Unidentified speaker: I know, I've seen it.
8:23
Unidentified speaker: The potential economic benefit is real, but the surrounding architecture determines whether that potential becomes an actual economic result.
8:36
Unidentified speaker: This is why economic architecture sits where it does inside the AI labor architecture framework.
8:45
Unidentified speaker: Not as a pillar, it sits as an outcome.
8:49
Unidentified speaker: It isn't where I want you to start.
8:53
Unidentified speaker: Because I don't think you should sit down and say, right, today we're going to redesign our economic architecture.
8:59
Unidentified speaker: What does that even mean?
9:02
Unidentified speaker: You can decide you want better margins.
9:04
Unidentified speaker: You can decide you want more capacity.
9:07
Unidentified speaker: You can decide you want to grow without increasing headcount at the same rate.
9:12
Unidentified speaker: You can decide you want greater consistency.
9:14
Unidentified speaker: Those are objectives.
9:16
Unidentified speaker: But you don't produce them by simply declaring the economic result you want.
9:23
Unidentified speaker: You produce them by changing the architecture underneath the economics.
9:29
Unidentified speaker: Start with the work.
9:32
Unidentified speaker: What outcomes actually need to be produced?
9:36
Unidentified speaker: What work is required to produce them?
9:39
Unidentified speaker: What work exists only because of the way the business operates today?
9:45
Unidentified speaker: What invisible work are people performing that you haven't formally recognized?
9:51
Unidentified speaker: Then look at responsibility.
9:54
Unidentified speaker: Who actually owns these outcomes?
9:56
Unidentified speaker: What are they allowed to do?
9:58
Unidentified speaker: What decisions can they make?
10:00
Unidentified speaker: What information can they access?
10:03
Unidentified speaker: Where does their responsibility begin and where does it end?
10:07
Unidentified speaker: Now I'm sorry to interrupt this amazing conversation, but I wanted to make sure you knew that smartphone, the only phone system built for real estate investors, connects to the best-in-class real estate investing CRMs.
10:21
Unidentified speaker: And because as it's my show and I am co-founder and chairman of Smartphone, I got some great deals for you.
10:27
Unidentified speaker: Click the link and make take advantage of 5,000 minutes free calling with smartphone, the only phone system built for real estate investors.
10:37
Unidentified speaker: Click and call.
10:40
Unidentified speaker: Close more leads.
10:42
Unidentified speaker: Back to the show.
10:43
Unidentified speaker: Look at execution.
10:45
Unidentified speaker: Who should do the work?
10:47
Unidentified speaker: A person?
10:49
Unidentified speaker: An AI agent?
10:51
Unidentified speaker: A system?
10:52
Unidentified speaker: Some combination of all three?
10:55
Unidentified speaker: Where should execution be shared?
10:58
Unidentified speaker: Where does one form of labor create more value than another?
11:02
Unidentified speaker: And then you need the infrastructure that allows that workforce to actually operate.
11:09
Unidentified speaker: How does the work arrive?
11:11
Unidentified speaker: How does it move?
11:12
Unidentified speaker: How does context travel with it?
11:14
Unidentified speaker: How does one worker hand responsibility into another?
11:19
Unidentified speaker: And what happens when circumstances change?
11:23
Unidentified speaker: How does the business know what should happen next?
11:27
Unidentified speaker: Now, once you start answering those questions, the economics begin to change as a consequence of the design.
11:36
Unidentified speaker: Go back to our 10-person customer success team.
11:39
Unidentified speaker: This time, don't start by saying, we want to get this down to three.
11:44
Unidentified speaker: Forget that number.
11:46
Unidentified speaker: Take the economics off the table for a moment.
11:50
Unidentified speaker: Instead, understand what those 10 people are actually doing.
11:54
Unidentified speaker: Now, maybe you discover that a significant amount of their time is spent repeatedly answering predictable customer questions.
12:03
Unidentified speaker: Okay, well, that work can potentially be executed differently.
12:08
Unidentified speaker: Maybe you discover they're manually monitoring customers to identify who might need attention.
12:14
Unidentified speaker: Well, an AI agent may be able to maintain that awareness continuously across every customer.
12:20
Unidentified speaker: Maybe people are spending hours every week recovering context from different systems before they can respond.
12:28
Unidentified speaker: That's an infrastructure architecture problem.
12:32
Unidentified speaker: Maybe your best customer success people are spending a huge amount of time chasing customers who haven't completed predictable actions.
12:41
Unidentified speaker: And maybe an AI agent can own that responsibility and execute it consistently across thousands of customers.
12:48
Unidentified speaker: And maybe some of the work shouldn't exist at all.
12:52
Unidentified speaker: Maybe it only exists because information doesn't move properly.
12:57
Unidentified speaker: Maybe it exists because responsibility is unclear, or because the process was designed around a limitation you no longer have.
13:07
Unidentified speaker: Now the business starts changing.
13:09
Unidentified speaker: Not because we decided to remove seven salaries.
13:13
Unidentified speaker: No, because we redesigned how customer success actually works.
13:18
Unidentified speaker: And once we've done that, maybe we discover something surprising.
13:22
Unidentified speaker: Maybe three people genuinely can support the same number of customers that previously required 10.
13:30
Unidentified speaker: Great.
13:31
Unidentified speaker: That is a real economic outcome.
13:35
Unidentified speaker: But maybe that's not actually the biggest opportunity.
13:38
Unidentified speaker: Maybe those same 10 people working alongside properly trained AI agents and inside designed responsibilities and infrastructure, maybe they can now support five times the number of customers.
13:53
Unidentified speaker: Maybe that's economically more valuable than reducing the team.
13:59
Unidentified speaker: Or maybe the right answer is somewhere in between.
14:02
Unidentified speaker: The point is this we didn't know the economic answer until we redesigned the business.
14:10
Unidentified speaker: And that's the mistake I want companies to avoid.
14:14
Unidentified speaker: Don't decide what AI should save you before you understand what AI allows you to redesign.
14:23
Unidentified speaker: Because those are two completely different ways of looking at the opportunity.
14:28
Unidentified speaker: One looks at your existing business and says, How can I make this cheaper?
14:33
Unidentified speaker: The other says, given the forms of labor now available to me, how should this business operate?
14:41
Unidentified speaker: And once you answer the second question, you might discover that the economics available to you are much more interesting than the savings you originally went looking for.
14:53
Unidentified speaker: Because this isn't just about payroll.
14:57
Unidentified speaker: Think about what starts happening when you redesign work properly.
15:02
Unidentified speaker: An AI agent may be able to execute work immediately instead of waiting for somebody to become available.
15:10
Unidentified speaker: That speed that changes.
15:22
Unidentified speaker: That changes consistency, doesn't it?
15:25
Unidentified speaker: And they may be able to maintain thousands of active contacts simultaneously.
15:30
Unidentified speaker: Now that changes your capacity.
15:36
Unidentified speaker: That changes availability.
15:39
Unidentified speaker: People may spend significantly more of their time performing the work where their contribution creates the greatest value.
15:48
Unidentified speaker: And that changes the economics of the people you already have.
15:54
Unidentified speaker: Invisible work may disappear because the infrastructure no longer requires people to compensate for it.
16:01
Unidentified speaker: That changes cost.
16:04
Unidentified speaker: And perhaps most importantly, the relationship between growth and headcount may start changing.
16:12
Unidentified speaker: Now, there has been a fairly stubborn relationship between the amount of work a business needs to execute and the number of people required to execute that work.
16:23
Unidentified speaker: Not perfectly.
16:24
Unidentified speaker: Technology has obviously improved productivity enormously, but in many parts of a business, growth still creates a labor demand.
16:35
Unidentified speaker: More customers, more support people, more opportunities, more salespeople, more transactions, more operations, more complexity, more managers, and more coordination.
16:45
Unidentified speaker: And eventually growth requires another layer of people to manage the people you added to handle the growth.
16:52
Unidentified speaker: AI labor gives us the opportunity to change some of those relationships, but only if the architecture allows us to.
17:01
Unidentified speaker: If every thousand new customer conversations create another thousand manual handoffs, or if every increase in AI execution creates more work for somebody downstream, if people still need to carry context between systems, or exceptions still pile up with one overwhelmed manager, or the business still relies on somebody remembering what needs to happen next, then eventually your economics run back into the architecture.
17:33
Unidentified speaker: That's why I think chasing the economic benefit first is so dangerous.
17:38
Unidentified speaker: You can optimize a spreadsheet before you've optimized the business.
17:44
Unidentified speaker: And those savings can look incredibly convincing right up until the underlying weaknesses start showing themselves.
17:51
Unidentified speaker: I would much rather build the foundation first.
17:56
Unidentified speaker: Understand the work, define the responsibilities, design the workforce, build the infrastructure, then look at what economics emerge.
18:06
Unidentified speaker: Because when those foundations are strong, the economic benefit isn't something you're constantly trying to force.
18:14
Unidentified speaker: It becomes structural.
18:16
Unidentified speaker: You're not getting a temporary productivity improvement because somebody found a clever way to use AI.
18:24
Unidentified speaker: The way the work is designed has changed.
18:27
Unidentified speaker: The way responsibility is allocated has changed.
18:30
Unidentified speaker: The way execution is distributed has changed.
18:35
Unidentified speaker: The way work moves has changed.
18:37
Unidentified speaker: The amount of capacity available to the business has changed.
18:41
Unidentified speaker: And therefore, the economics of producing the outcome have changed.
18:46
Unidentified speaker: That's what I mean by economic architecture.
18:50
Unidentified speaker: And I think this distinction matters because this could be an enormous pressure over the next few years to turn AI into a financial calculation.
19:00
Unidentified speaker: How many jobs can it replace?
19:02
Unidentified speaker: How much can we save?
19:04
Unidentified speaker: How much more can we produce?
19:05
Unidentified speaker: What's the ROI?
19:07
Unidentified speaker: Now, those questions aren't wrong.
19:10
Unidentified speaker: They're just not where I think you should start.
19:14
Unidentified speaker: Because if you start with where can we cut people, you're designing around a cost target.
19:22
Unidentified speaker: If you start with how much more can this AI agent do, you're designing around a capability.
19:30
Unidentified speaker: But if you start with how should this business actually operate now that I have multiple forms of labor available to me, you're designing the business.
19:41
Unidentified speaker: And then you can measure the economics of what you've built.
19:46
Unidentified speaker: So here's something I'd encourage you to do inside your own business.
19:50
Unidentified speaker: Think about one place where you've already looked at AI and done the math.
19:55
Unidentified speaker: Maybe you've thought, we could save a load of money here, or or we could do three times as much without hiring anybody.
20:03
Unidentified speaker: Don't throw that calculation away.
20:06
Unidentified speaker: Just put it to one side.
20:08
Unidentified speaker: Then go underneath it.
20:10
Unidentified speaker: What work are you actually trying to produce?
20:13
Unidentified speaker: What responsibilities exist?
20:16
Unidentified speaker: What are the people currently involved actually doing?
20:20
Unidentified speaker: What invisible work exists around it?
20:24
Unidentified speaker: What should be executed by people?
20:26
Unidentified speaker: What should be executed by AI agents?
20:29
Unidentified speaker: What should systems handle?
20:31
Unidentified speaker: How should that execution move between them?
20:34
Unidentified speaker: And what infrastructure would need to change?
20:37
Unidentified speaker: And only when you've redesigned that picture, then come back to the spreadsheet and ask, what are the economics now?
20:47
Unidentified speaker: I suspect in some cases, you'll discover the savings that you originally imagined wasn't realistic.
20:55
Unidentified speaker: But in others, I think you'll discover something much better and much more interesting.
21:00
Unidentified speaker: The economic opportunity was significantly bigger than the one you initially calculated.
21:06
Unidentified speaker: You just couldn't see it while you were looking at the old business.
21:11
Unidentified speaker: So that's why economic architecture comes at the end of the framework.
21:16
Unidentified speaker: Not because the economics matter less, quite the opposite.
21:20
Unidentified speaker: Because if you want to maximize the economic impact of AI labor, I think you have to earn your way there through the architecture.
21:30
Unidentified speaker: Start with the economics, and the foundations underneath your assumptions may be incredibly shaky.
21:37
Unidentified speaker: But build the foundations first.
21:39
Unidentified speaker: The economics become an outcome of the business you've designed.
21:44
Unidentified speaker: And that's when AI stops simply making your existing business a little cheaper or a little faster.
21:52
Unidentified speaker: It starts changing the economics of what that business is capable of becoming.com to check out all episodes and get special discounts on tons of awesome real estate technology platforms.
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3 and interpreted by Qwen3.6-35B-A3B-GGUF:UD-Q4_K_XL. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights